feat: same-turn memory retrieval in provider mode #6

2026-08-15T17:52:36Z

ARIA commented

2026-08-15 17:52:36 +00:00

In provider mode, prefetch() returned the previous turn's cached result, so memory context was always one turn behind. With ~40ms local retrieval there's no reason for the lag.

Changes:

prefetch(query) reuses the in-flight background search when it matches the current query, else searches synchronously — same-turn injection
_prefetch_query scopes cached results to their query; result is reset when a new search starts and thread writes are guarded, preventing stale cross-turn injection after a timed-out join
Sync prefetch capped at 3s total (split across both requests in case-insensitive mode) so a hung server can't stall the LLM hot path for 10s
LocalMem0Client gains optional per-request timeout override
AGENTS.md / README updated to match

Verified: py_compile passes; two review passes, all findings addressed.

In provider mode, prefetch() returned the previous turn's cached result, so memory context was always one turn behind. With ~40ms local retrieval there's no reason for the lag. Changes: - prefetch(query) reuses the in-flight background search when it matches the current query, else searches synchronously — same-turn injection - _prefetch_query scopes cached results to their query; result is reset when a new search starts and thread writes are guarded, preventing stale cross-turn injection after a timed-out join - Sync prefetch capped at 3s total (split across both requests in case-insensitive mode) so a hung server can't stall the LLM hot path for 10s - LocalMem0Client gains optional per-request timeout override - AGENTS.md / README updated to match Verified: py_compile passes; two review passes, all findings addressed.

ARIA added 1 commit 2026-08-15 17:52:37 +00:00

feat: same-turn memory retrieval in provider mode c79cf43ff8

prefetch() now reuses the in-flight background search when it matches
the current query, or falls back to a synchronous search (capped at 3s,
split across both requests in case-insensitive mode) so memories are
injected on the same turn instead of one turn behind.

- Track _prefetch_query to scope cached results to their query
- Reset _prefetch_result when starting a new background search to
  prevent stale cross-turn injection after a timed-out join
- Guard background thread writes against superseded queries
- Add optional per-request timeout override to LocalMem0Client
- Update AGENTS.md and README to describe the same-turn flow

Zephyre_Review commented

2026-08-15 17:52:41 +00:00

✅ No issues found — changes look consistent with the stated intent. Ready to be merged.

Pakobbix merged commit 185eec75f4 into main

2026-08-15 17:53:03 +00:00

Pakobbix deleted branch feature/same-turn-prefetch

2026-08-15 17:53:04 +00:00

Sign in to join this conversation.

2 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: ARIA/mem0-local-hermes-plugin#6